The IMDI Metadata Set, its Tools and accessible Linguistic Databases

Daan Broeder, Freddy Offenga, Don Willems, Peter Wittenburg
Max-Planck-Institute for Psycholinguistics
broeder@mpi.nl

The ISLE Metadata Initiative (IMDI) was founded in February 2000 and has
resulted in a metadata set for multimedia/multimodal language resources, in
a user-friendly editor, a browser supporting an interactive search and
browse mode, several converter utilities to facilitate the step over from
legacy metadata, and a showcase in which 8 European institutions
participated. IMDI is currently used by some large projects in Europe to
organize their archive of language resources and to make the data more
easily accessible.

The demonstration will explain the definition of the metadata elements [1],
the closed and open vocabularies for the elements, the corresponding
constraints, and their usage based on concrete examples from field
linguistics and language engineering. It will also describe the session
concept, discuss the IMDI to OLAC mapping [2], and demonstrate which
information will be lost during the OLAC harvesting step.

The metadata editor [3] supports the IMDI set and offers a user-friendly
interface to allow users to enter IMDI like descriptions. Since the IMDI
set is built up on blocks of descriptor elements, the editor was built in
such a way that recurring blocks of information can be reused, i.e. the
user can store and reuse his own set of descriptor blocks. The editor also
supports the usage of external vocabulary sets by contacting the
corresponding URLs. A number of very useful converters and utilities were
built to efficiently transfer legacy data and to increase the operational
efficiency. The major points will be explained.

The IMDI BCBrowser [4] supports browsing in complex and distributed
hierarchies of metadata descriptions where nodes can and will be associated
with many types of linguistic descriptions such as field matters. All
resources, whether metadata descriptions or linguistic notes, can be linked
by specifying URLs. The user can switch between searching and browsing to
give a maximum of navigation efficiency.  The browser also allows the user
to directly start suitable tools when resources have been found.

Metadata search is implemented on the basis of text search in optimised
text files so that there is no requirement for database software, an
approach that is at the present time sufficient for corpora of the size we
use.

Finally, the IMDI showcase [5] will be demonstrated which shows the
IMDI-BCBrowser operating in a distributed environment containing metadata
from 8 European institutions. The showcase supports multiple hierarchies
and a geographic approach by using a clickable world map to point to
institutions and languages. The demonstration will end with a critical
evaluation of what has been done so far.

References:
[1] IMDI metadata set http://www.mpi.nl/ISLE_MetaData_2.5.pdf 
[2] IMDI to OLAC Mapping (to be published)
[3] IMDI BC-Editor IMDI-BCEditor, http://www.mpi.nl/ISLE/tools
[4] IMDIBCBrowser, IMDI-BCEditor, http://www.mpi.nl/ISLE/tools[5] IMDI Showcase,
[5] http://www.mpi.nl/world/data/LUNDDEMO/data/imdi_demo.html